PHYSICAL REVIEW D, VOLUME 62, 114506

Perturbative renormalization of weak Hamiltonian four-fermion operators with overlap fermions

Stefano Capitani*
Center for Theoretical Physics, Laboratory for Nuclear Science, Massachusetts Institute of Technology, 77 Massachusetts Avenue,
Cambridge, Massachusetts 02139

Leonardo Giusti†
Department of Physics, Boston University, 590 Commonwealth Avenue, Boston, Massachusetts 02215
共Received 17 July 2000; published 30 October 2000兲
The renormalization of the most general dimension-six four-fermion operators without power subtractions is
studied at one loop in lattice perturbation theory using overlap fermions. As expected, operators with different
chirality do not mix among themselves and parity-conserving and parity-violating multiplets renormalize in the
same way. The renormalization constants of unimproved and improved operators are also the same. These
mixing factors are necessary to determine the physical matrix elements relevant to many phenomenological
applications of weak interactions. The most important are the K 0 -K̄ 0 and B 0 -B̄ 0 mixings in the standard model
and beyond, the ⌬I⫽1/2 rule and ⑀ ⬘ / ⑀ .
PACS number共s兲: 11.15.Ha, 12.38.Gc, 14.40.Aq

I. INTRODUCTION To compute non-leptonic weak matrix elements it is es-

Since the original proposals of using lattice QCD to study
hadronic weak decays 关1兴, substantial theoretical and nu-
merical progress has been made. In the most popular lattice
regularizations, i.e. Wilson and staggered fermions, the main
theoretical aspects of the renormalization of composite four-
fermion operators are fully understood 关2,3兴. The calcula-
tions of the matrix elements relevant for the K 0 -K̄ 0 and B 0 -
B̄ 0 mixings have reached a level of accuracy which is un-
matched by any other approach 关4,5兴; increasing precision
has also been achieved in determining the electroweak pen-
guin amplitudes necessary for the prediction of the
C P-violation parameter ⑀ ⬘ / ⑀ 关6–12兴. Finally, matrix ele-
ments of ⌬S⫽2 operators, relevant to the study of flavor
changing neutral current 共FCNC兲 effects in several exten-
sions of the standard model 共supersymmetry, left-right sym-
metric models, multi-Higgs models, etc.兲, have also been
computed 关11–13兴. Some of these lattice predictions have
been fundamental in constraining the parameters of the
Cabibbo-Kobayashi-Maskawa 共CKM兲 matrix in the standard
model 共SM兲 关14,15兴 and beyond 关16兴.
Nevertheless, some fundamental phenomena, such as the
⌬I⫽1/2 rule in non-leptonic kaon decays and the value of
⑀ ⬘ / ⑀ , which measures the direct C P violation in kaon de-
cays, are far from being understood. In the standard model
these effects can be explained only if the non-perturbative
physics gives contributions to matrix elements definitively
larger than their factorized values 关15兴. Therefore a non-
perturbative determination of the relevant matrix elements is
crucial for predicting these quantities. Lattice QCD is the
only method which can address these problems from first
principles. Techniques have been developed for both Wilson
and staggered fermions, but these methods have not yet pro-
duced useful results 关6,17兴.
*Email address: stefano@mitlns.mit.edu
†
Email address: lgiusti@bu.edu
sential to construct renormalized operators in definite chiral
representations. In the Wilson or clover-improved lattice
regularizations, bare operators do not have a definite chiral
behavior due to the presence of the symmetry breaking term
in the action. Renormalized operators with the correct chiral
properties are recovered as linear combinations of operators
with different chirality. The standard model O ⌬S⫽2 operator
is the most popular example: its matrix element between
pseudoscalar states should vanish as M ␲2 in the limit M ␲
→0, while the matrix elements of the wrong-chirality opera-
tors which mix with it are expected to go to a constant 共and
in fact in the kaon mass region they are 2–10 times larger
than the SM one兲. Therefore the correct chiral behavior of
O ⌬S⫽2 is recovered only if the finite mixing constants are
known with high precision. This has been a long-standing
problem and has only been solved using non-perturbative
共NP兲 renormalization techniques 关9,10,18兴. The situation be-
comes even worse when the lack of chiral symmetry compli-
cates mixings with lower dimensional operators. This is one
of the obstacles to a reliable computation of the matrix ele-
ments relevant for ⌬I⫽1/2 and ⑀ ⬘ / ⑀ . On the other hand,
with staggered fermions chiral symmetry is preserved, but
solving the doubling problem and defining operators with the
correct flavor and spin quantum numbers is far from trivial.
Only recently has it been understood 关19–23兴 that chiral
and flavor symmetries can be preserved simultaneously on
the lattice, without fermion doubling, if the fermionic opera-
tor D satisfies the Ginsparg-Wilson relation 共GWR兲 关24兴
a
␥ 5 D⫹D ␥ 5 ⫽ D ␥ 5 D. 共1兲
␳
The GWR implies an exact symmetry of the fermion action
at non-zero lattice spacing, which may be regarded as a lat-
tice form of the standard chiral rotation 关23兴. Nevertheless, it
is important to stress that, locality, the absence of doubler
modes and the correct classical continuum limit are not guar-
anteed by the GWR in Eq. 共1兲. Indeed, there exist lattice

0556-2821/2000/62共11兲/114506共14兲/$15.00 62 114506-1 ©2000 The American Physical Society

STEFANO CAPITANI AND LEONARDO GIUSTI
fermion actions which satisfy the GWR but which do not
meet the above requirements 关25兴. A breakthrough in this
field was achieved through the domain-wall formulation of
lattice fermions 关26兴 and by Neuberger through the overlap
formulation 关21兴. He found a solution of the GWR which
satisfies all of the above requirements and is local1 关27兴
D N⫽
␳
a 冉
1⫹X
1
冑X † X 冊 The renormalization constants we have computed are also
1 key observation in these cases 关31兴 is that the lower-
X⫽D W ⫺ ␳ ,
a dimension operators do not change the anomalous dimen-
where D W is the Wilson-Dirac operator and 0⬍ ␳ ⬍2r 共see
below兲. A further remarkable property of Ginsparg-Wilson
fermions is the absence of O(a) discretization errors in the
action and therefore in the spectrum of the theory. Neverthe-
less, the local fermionic operators have to be improved to
remove the O(a) effects in the matrix elements. This step is
greatly simplified by the Ginsparg-Wilson relation, as it al-
lows the construction of O(a)-improved operators to all or-
ders in g 2 关28兴, which renormalize with the same renormal-
ization factors of the unimproved ones. On the other hand,
the complicated form of the Neuberger operator in Eq. 共2兲
renders its numerical implementation quite demanding for
the present generation of computers. However, some
progress has been achieved 关29兴 and Monte Carlo simula-
tions seem to be already feasible, at least in the quenched
approximation.
Once the action satisfies the GWR and all the properties
described above, the quark masses renormalize only multi-
plicatively and mixings of operators with different chirality
are forbidden even at finite cutoff 关22兴. The most general set
of dimension-six four-fermion operators without power sub-
tractions is
O ⌫ A ⌫ B ⫽ ¯␺ 1 ⌫ Al ␺ 2 • ¯␺ 3 ⌫ Bl ␺ 4
O ⌫F ⫽ ¯␺ 1 ⌫ Al ␺ 4 • ¯␺ 3 ⌫ Bl ␺ 2 .
A⌫B
The main result of this paper is the evaluation of the mixing
pattern of these operators at one loop in perturbation theory
in the overlap lattice regularization: we compute their renor-
malization constants and we show how the Neuberger action
greatly simplifies the mixing among them.
The renormalization constants are necessary for extracting
physical matrix elements from numerical simulations. The
matrix elements of the operators studied in this work are
relevant in many phenomenological applications of weak in-
teractions. They are necessary for predicting the K 0 -K̄ 0 and
B 0 -B̄ 0 mixing amplitudes in the standard model and in sev-
1
From Eq. 共2兲 it is clear that the overlap operator is not ultra-local.
The Neuberger kernel satisfies a more general definition of locality;
i.e., it is exponentially suppressed at large distances with a decay
rate proportional to 1/a.
PHYSICAL REVIEW D 62 114506
eral extensions of it 共supersymmetry, left-right symmetric
models, multi-Higgs models, etc.兲. By using chiral perturba-
tion theory, they can also be related to the ⌬I⫽3/2 matrix
elements relevant for the prediction of ⑀ ⬘ / ⑀ . They are also
necessary for estimating the B s0 -B̄ s0 width difference and the
O(1/m 3b ) corrections in inclusive b-hadron decay rates 关30兴.
relevant for the study of the ⌬I⫽1/2 rule and ⑀ ⬘ / ⑀ on the
lattice. The operators involved in these computations can
mix with same- as well as lower-dimension operators. The
共2兲
sions of the original operators. Therefore the mixings among
dimension-six operators can be obtained from those of the
operators studied in this paper.
Perturbative computations with the overlap Dirac operator
are much more cumbersome than for Wilson fermions, due
to the complicated structure of the vertices in the action 共see
Appendix A兲. For this reason it is very useful to use an
intermediate renormalization scheme to separate as much as
possible the computations performed using lattice perturba-
tion theory 共where, for example, the Fierz rearrangements
work兲 from the continuum perturbation theory which is
much simpler. The regularization independent 共RI兲 renormal-
ization scheme proposed in 关32,33兴 is the optimum choice:
we first renormalize lattice operators in the RI scheme using
only lattice perturbation theory and then the matching with
one of the modified minimal subtraction (MS) schemes is
done in continuum perturbation theory. Lattice and con-
tinuum regularizations are used independently taking full ad-
vantage of their properties.
In Ref. 关34兴, renormalization constants of local bilinear
fermion operators have been computed in the MS renormal-
ization scheme using the lattice overlap regularization. We
computed these renormalization constants in the RI scheme,
matched them in MS and verified that our results are in
agreement with those of Ref. 关34兴. Then we computed the
共3兲 renormalization constants of the four-fermion operators we
are interested in following two independent ways: direct
computation, such as in Ref. 关35兴, and one which uses the
Fierz and charge conjugation rearrangements to compute the
four-fermion Green’s functions in terms of the bilinear ones
关36,8兴. The agreement of the results is also a non-trivial
check that the numerical integrals with the Neuberger propa-
gators and vertices are well estimated.
A further determination of the renormalization constants
considered in this paper can be obtained using numerical NP
methods such as those in Refs. 关33,39兴, applied to Neuberger
fermions. However, perturbative estimates are very useful
because often they are very good approximations and they
furnish a consistency check of the NP methods. Moreover,
for Neuberger fermions the perturbative computations can
remain the only determinations of the renormalization con-
stants for some time.
The paper is organized as follows: in Sec. II we define the
overlap fermion action used, in Sec. III we discuss the bilin-
ear renormalization constants, in Sec. IV we address the mix-
114506-2
PERTURBATIVE RENORMALIZATION OF WEAK . . . PHYSICAL REVIEW D 62 114506

ing of the four-fermion operators and report the results for independent renormalization schemes 共RI and MS) and
the renormalization constants, and in Sec. V we state our therefore all our computations are performed with massless
conclusions. quarks.

II. BASIC DEFINITIONS III. RENORMALIZATION OF BILINEAR OPERATORS

The QCD lattice regularization we consider for massless In this section we set our notation for bilinear operators
fermions is described by the action and we report the results we have obtained for their renor-

冋 册
malization constants in the RI and MS schemes.
6 1
S L⫽
g 20
兺P 1⫺ Tr关 U P ⫹U †P 兴
6
A generic non-singlet quark bilinear is defined as

O ⌫ 共 x 兲 ⫽ ¯␺ 1 共 x 兲 ⌫ ␺ 2 共 x 兲 共9兲
Nf

⫹ 兺 兺 ¯␺ i共 x 兲 D N共 x,y 兲 ␺ i共 y 兲 ,
i⫽1 x,y
where, in standard notation, U P is the Wilson plaquette and
g 0 the bare coupling constant. D N is the Neuberger-Dirac
operator defined in Eq. 共2兲, where the massless Wilson op-
erator D W is defined as
1 r
D W ⫽ ␥ ␮ 共 ⵜ ␮ ⫹ⵜ ␮* 兲 ⫺ aⵜ ␮* ⵜ ␮
2 2
共4兲 where the flavors ␺ 1 ⫺ ␺ 2 are different and ⌫ is a generic
Dirac matrix. The GWR ensures that no lattice artifacts of
O(a) are present in the action and therefore also not in the
spectrum of the theory. However, matrix elements of opera-
tors are still affected by O(a) discretization errors that have
to be removed by improving the fermionic operators. In
Refs. 关28兴 it is shown that, for massless quarks, the improved
bilinear operator is given by
共5兲

and ⵜ ␮ and ⵜ ␮* are the forward and backward lattice deriva-

冉
O ⌫I ⫽ ¯␺ 1⫺
1
2␳
aD N ⌫ 1⫺ 冊 冉
1
2␳
aD N ␺ . 冊 共10兲

tives, i.e.
It can be proved 关34,28兴 that the renormalization constants of
1 O ⌫I are the same as those of the corresponding O ⌫ . There-
ⵜ ␮ ␺ 共 x 兲 ⫽ 关 U ␮ 共 x 兲 ␺ 共 x⫹a ␮
ˆ 兲 ⫺ ␺ 共 x 兲兴 fore all the results obtained in this section for local bilinear
a operators also remain valid for the corresponding improved
operators.
1 The renormalized bilinear operators are defined by
ⵜ ␮* ␺ 共 x 兲 ⫽ 关 ␺ 共 x 兲 ⫺U ␮† 共 x⫺a ␮
ˆ 兲 ␺ 共 x⫺a ␮
ˆ 兲兴 .
a
共6兲 Ô ⌫ 共 ␮ 兲 ⫽Z ⌫ 共 a ␮ 兲 O ⌫ 共 a 兲 共11兲
2
The range of the Wilson parameter is 0⬍r⭐1 and U ␮ (x) is and the chiral symmetry in Eq. 共7兲 imposes the constraints
the lattice gauge link. Equation 共1兲 implies a continuous
symmetry of the fermion action in Eq. 共4兲, which be may Z A ⫽Z V
interpreted as a lattice form of chiral symmetry 关23兴:

冉 冊 冉 冊
Z P ⫽Z S 共12兲
1 1
␦␺⫽␥5 1⫺ aD N ␺ , ␦ ¯␺ ⫽ ¯␺ 1⫺ aD N ␥ 5 .
2␳ 2␳ on the renormalization constants.3
共7兲 The quark propagator in momentum space is denoted by
S(p) 共for the conventions adopted throughout the paper see
The corresponding flavor non-singlet chiral transformations Appendix B兲. The two-point fermionic Green’s function of a
are defined including a flavor group generator in Eq. 共7兲. The bilinear inserted at the origin (x⫽0) is
generalization to massive fermions is simple 关40兴: in Eq. 共4兲,
D N has to be replaced by G ⌫ 共 x 1 ,x 2 兲 ⫽ 具 ␺ 1 共 x 1 兲 O ⌫ 共 0 兲 ¯␺ 2 共 x 2 兲 典 , 共13兲

D N → 1⫺ 冉 1
2␳ 冊
am i0 D N ⫹m i0 共8兲 its Fourier transform with equal external momenta is

where m i0 is the bare physical quark mass of flavor i. The G ⌫共 p 兲 ⫽ 冕 dx 1 dx 2 e ⫺ip(x 1 ⫺x 2 ) G ⌫ 共 x 1 ,x 2 兲 , 共14兲
Feynman rules of the action defined in Eq. 共4兲 are given in
Appendix A and are in agreement with those used by and the corresponding amputated correlation function is de-
关34,41,42兴. Throughout this paper we will use only mass fined as

2 3
All the computations reported in this paper are performed with If the conserved axial and vector currents corresponding to the
r⫽1. chiral symmetry in Eq. 共7兲 were used, then Z V ⫽Z A ⫽1 关19兴.

114506-3
STEFANO CAPITANI AND LEONARDO GIUSTI PHYSICAL REVIEW D 62 114506

⌳ ⌫ 共 p 兲 ⫽S ⫺1 共 p 兲 G ⌫ 共 p 兲 S ⫺1 共 p 兲 . 共15兲 TABLE I. Perturbative values in the continuum and on the lat-

tice in the Landau gauge, where ¯␥ (0) ⫽ ␥ (0) /C F while the B’s con-
The renormalized quark propagator is tain the proper vertex contribution only.

␺ S V T
Ŝ 共 p, ␮ 兲 ⫽Z ␺ 共 a ␮ 兲 S 共 p,a 兲 , 共16兲
Continuum PT
where Z ␺1/2 is the quark field renormalization constant, and
¯␥ (0) 0 ⫺6 0 2
the renormalized Green’s functions are
⌬ MS,RI 0 ⫺4 0 0
Ĝ ⌫ 共 p, ␮ 兲 ⫽Z ␺ 共 ␮ a 兲 Z ⌫ 共 ␮ a 兲 G ⌫ 共 p,a 兲 , 共17兲 ␳ Lattice PT

ˆ ⌫ 共 p, ␮ 兲 ⫽Z ␺⫺1 共 ␮ a 兲 Z ⌫ 共 ␮ a 兲 ⌳ ⌫ 共 p,a 兲 . 0.2 ⫺235.80762 1.31942 1.52122 1.58848

⌳
0.3 ⫺150.61868 1.89625 1.52277 1.39828
共18兲
0.4 ⫺108.19798 2.38060 1.52448 1.23911
The anomalous dimensions of composite operators are de- 0.5 ⫺82.86081 2.80522 1.52637 1.10009
fined as 0.6 ⫺66.05227 3.18782 1.52845 0.97532
0.7 ⫺54.10921 3.53927 1.53074 0.86124

冉 冊 ⫺45.20179
⬁ 0.8 3.86686 1.53329 0.75543
d ␣s i⫹1
␥ ⌫ 共 ␣ s 兲 ⫽⫺2Z ⌫⫺1 ␮ 2
d␮
Z ⫽
2 ⌫ 兺
i⫽0
␥ ⌫(i)
4␲
, 共19兲 0.9 ⫺38.31447 4.17577 1.53611 0.65622
B 1.0 ⫺32.83862 4.46989 1.53924 0.56236
1.1 ⫺28.38734 4.75224 1.54274 0.47290
where ␣ s ⫽g 2 /4␲ is the strong coupling constant. If a renor- 1.2 ⫺24.70304 5.02527 1.54665 0.38711
malization scheme which preserves the chiral Ward identities 1.3 ⫺21.60760 5.29104 1.55105 0.30438
共WIs兲 is chosen 共such as MS or RI兲, then 1.4 ⫺18.97397 5.55135 1.55601 0.22422
1.5 ⫺16.70910 5.80783 1.56163 0.14623
␥ A ⫽ ␥ V ⫽0, 1.6 ⫺14.74330 6.06201 1.56804 0.07005
1.7 ⫺13.02336 6.31544 1.57541 -0.00460
␥ P ⫽ ␥ S ⫽⫺ ␥ m , 共20兲 1.8 ⫺11.50798 6.56970 1.58394 -0.07798

where ␥ m is the anomalous dimension of the quark mass.

Moreover, the ␥ ⌫(0) s are renormalization-scheme independent
and gauge invariant and their values are reported in Table I.
The anomalous dimensions are properties of the renor-
malized operators: they are independent of the regularization
scheme adopted and depend only on the renormalization
conditions imposed 共the renormalization scheme兲 and the
renormalized coupling constant used. For a given operator,
defining the renormalization conditions is equivalent to fix-
ing ␥ ⌫ ( ␣ s ); i.e., ␥ ⌫ ( ␣ s ) uniquely determines the scheme in
a regularization invariant way. Therefore, ␥ ⌫ can be com-
puted in a simpler regularization, such as naive dimensional
regularization 共NDR兲, provided that the renormalization con-
ditions are the same. The evolution of the renormalized op-
erators is determined by solving the renormalization group
equations and, in a given scheme, is
c ⌫共 ␮ 兲
Ô ⌫ 共 ␮ 兲 ⫽ O 共 ␮ 兲,
c ⌫共 ␮ 0 兲 ⌫ 0
where c ⌫ ( ␮ ), at next-to-leading order 共NLO兲,
(0)
c ⌫ 共 ␮ 兲 ⫽ 关 ␣ s 共 ␮ 兲兴 ␥ ⌫ /(2 ␤ 0 )
冋 1⫹
4␲
␣ s 共 ␮ 兲 ␥ ⌫(1) ␤ 1 ␥ ⌫(0)
is scheme dependent. In Eq. 共22兲, ␤ 0 and ␤ 1 are the first two
coefficients of the ␤ function of QCD. For ␥ ⌫(1) in the RI or
MS scheme see for example 关43兴. The renormalization group
invariant 共RGI兲 operators can be defined as 关43–45兴
1
Ô ⌫RGI⫽ Ô 共 ␮ 兲 ⫽Z ⌫RGI共 a 兲 O ⌫ 共 a 兲 , 共23兲
c ⌫共 ␮ 兲 ⌫
where
Z ⌫共 ␮ a 兲
Z ⌫RGI共 a 兲 ⫽ . 共24兲
c ⌫共 ␮ 兲
Ô ⌫RGI is independent of the scheme, scale and gauge chosen
to renormalize the operators, while Z ⌫RGI(a) depends only on
the regularization and not on the scheme, scale and gauge.
As discussed in the Introduction, it is very helpful to sepa-
rate computations which can be performed using only lattice
共21兲
perturbation theory from those done in the continuum. In this
regard the RI renormalization scheme proposed in 关32,33兴 is
the optimum choice: it preserves all the relevant symmetries
共chirality and switch, see below兲 and it can also be easily
冉
2␤0
⫺
2 ␤ 20
冊册 ,
implemented non-perturbatively 关33兴. The matching with a
‘‘continuum’’ renormalization scheme, for example MS, re-
mains necessary because almost all the Wilson coefficients
共22兲 are computed in the continuum, but it can be done using
continuum perturbation theory only. Therefore lattice and
continuum regularizations are used independently, taking full
advantage of their properties. In the following we will indi-

114506-4
PERTURBATIVE RENORMALIZATION OF WEAK . . . PHYSICAL REVIEW D 62 114506

cate the renormalization scheme adopted with a superscript c ⌫␹ 共 ␮ 兲

in the renormalization quantities, i.e. Z ⌫RI and Z MS
⌫ for RI and Z ⌫␹ 共 ␮ a 兲 ⫽c ⌫␹ 共 ␮ 兲 Z ⌫RGI共 a 兲 ⫽ Z ⌫RI共 ␮ a 兲 , 共28兲
c ⌫RI共 ␮ 兲
MS schemes respectively.
Using the Feynman rules defined in Appendix A, we com-
where the c ⌫RI( ␮ ) are reported in 关43兴 for the bilinears, de-
puted the self-energy of the quark propagator and the ampu-
tated Green’s functions of bilinear operators between off- fined with the same conventions adopted in this paper.
shell quark states at one loop in perturbation theory in a To compare our results with 关34兴, we report explicitly the
generic covariant gauge. The full expressions we obtained matching coefficients for the MS scheme. The chiral WIs
are reported in Appendix D and they can be used to impose imply Z VRI⫽Z MS
V , Z A ⫽Z A and
RI MS

any renormalization condition in any covariant gauge. The

RI scheme is imposed by taking the trace of the amputated Z RI
S Z MS
S
Green’s functions in the Landau gauge ( ␣ ⫽0) 关33兴. The ⫽ ⫽1. 共29兲
renormalization constant of the quark field is then defined as Z RI
P Z MS
P

1 ⳵ S ⫺1 共 p 兲 A straightforward computation in NDR gives

Z ␺RI⫽⫺i Tr␥ ␮ 兩 2 2, 共25兲
⳵ p ␮ p ⫽␮
冋 册
48
␣ s MS,RI
⌫ 共 a ␮ 兲 ⫽ 1⫺C F
Z MS 共⌬ ⫹⌬ MS,RI
␺ 兲 Z ⌫RI共 a ␮ 兲 ,
while the renormalization constants of the bilinear operators 4␲ ⌫
are given by 共30兲

关 Z ␺RI共 ␮ a 兲兴 ⫺1 Z ⌫RI共 ␮ a 兲 Tr关 P ⌫ ⌳ ⌫ 共 p,a 兲兴 兩 p 2 ⫽ ␮ 2 ⫽1, 共26兲
where the trace is over both color and spin indices and P ⌫
are suitable projectors on the tree-level operators defined in
Appendix C. In general, these renormalization conditions
differ from the standard momentum subtraction (MOM) per-
turbative prescriptions 共see for example 关36兴兲, where the
renormalization constants are extracted only from terms pro-
portional to the tree-level matrix elements of the operators
under consideration. The differences are evident from the
expressions of the quark propagator and the amputated
Green’s functions reported in Appendix D. In the standard
procedure, finite terms such as p” p ␮ /p 2 in the vector current
are considered as part of the matrix elements of the opera-
tors, whereas with the projectors they give additional finite
contributions to the renormalization constants.4 These terms
are the same on the lattice and in the continuum and they
cancel when one computes the difference between the con-
tinuum and the lattice renormalization constants, as in the
standard perturbative procedure.
Using Eqs. 共25兲 and 共26兲 and the results reported in Ap-
pendix D we obtain
where the ⌬ MS,RI’s are reported in Table I. In the MS
scheme, the renormalization constants of gauge-invariant op-
erators are independent of the gauge. Therefore, the depen-
dence on the gauge and external states of the RI scheme
cancels the corresponding dependence of the matching coef-
ficients. Our results in MS are in agreement with those of
Ref. 关34兴.
IV. FOUR-FERMION OPERATORS
In this section we introduce the most general set of
dimension-six four-fermion operators we are interested in;
we analyze their mixing pattern, exploiting only the symme-
tries of the underlying theory; and finally we compute the
necessary renormalization constants at one loop in perturba-
tion theory. A very exhaustive non-perturbative analysis for
Wilson fermions has been done in Ref. 关18兴. We will pro-
ceed on the same lines with the advantage of having the
additional chiral symmetry in Eq. 共7兲 which forbids mixing
among operators which belong to multiplets with different
chirality 关22兴.
The generic set of four-fermion operators we are inter-

冋 册
ested in is
g 20 ¯␥ ␺(0)
Z ␺RI⫽1⫺ C
2 F
log共 ␮ a 兲 2 ⫹B ␺ O ⌫ A ⌫ B ⫽ ¯␺ 1 ⌫ Al ␺ 2 • ¯␺ 3 ⌫ Bl ␺ 4
16␲ 2

Z ⌫RI⫽1⫺
g 20
16␲
C
2 F
冋 ¯␥ ⌫(0)
2
log共 ␮ a 兲 ⫹B ⌫ ⫹B ␺ ,
2
册 O ⌫F
A⌫B
⫽ ¯␺ 1 ⌫ Al ␺ 4 • ¯␺ 3 ⌫ Bl ␺ 2 ,

where the flavors ␺ 1 – ␺ 4 are all different and the ⌫ Al are

共31兲

共27兲
where the ¯␥ ⌫(0) and B’s are reported in Table I.
Once the operators have been renormalized in RI, the
matching with another given scheme ␹ is obtained with
4 matrix elements can be removed by using the improved op-
Note that in the Landau gauge these terms are absent both in the
bilinears and in the four-fermion operators.
generic Dirac matrices with l representing the contracted Lo-
renz indices 共if any兲. The 20 operators in Eqs. 共31兲 form a
complete basis of four-fermion operators. Operators with
color indices contracted in a different way can be expressed
as linear combinations of the ones in Eq. 共31兲 using the color
Fierz identity 共B3兲 in Appendix B.
The O(a) discretization errors in on-shell four-fermion
erator

114506-5
STEFANO CAPITANI AND LEONARDO GIUSTI PHYSICAL REVIEW D 62 114506

冉
¯␺ 1⫺
1
2␳ 冊 冉
aD N ⌫ A 1⫺
1
2␳ 冊
aD N ␺ • ¯␺
and, for the parity-violating (P⫽⫺1) operators,

冉 冊 冉 冊
Q⫾ ⫾ ⫾
1 ⫽O VA ⫹O AV
1 1
⫻ 1⫺ aD N ⌫ B 1⫺ aD N ␺ . 共32兲
2␳ 2␳
Q⫾ ⫾ ⫾
2 ⫽O VA ⫺O AV
This can be proved along the lines used in Ref. 关28兴 for the
case of two-quark operators. Here we just sketch the argu-
ment, which is valid for any reasonable action which satisfies Q⫾ ⫾ ⫾
3 ⫽⫺O S P ⫹O PS
the GWR. Starting from the Neuberger operator D N , one can
define the associated operator 关25兴
Q⫾ ⫾ ⫾

冉 冊
⫺1 4 ⫽O S P ⫹O PS
1
K N ⫽ 1⫺ aD N DN , 共33兲
2␳ ⫾
Q⫾
5 ⫽O TT̃ , 共36兲
which has the same chiral properties as the continuum Dirac
operator, i.e. 兵 K N , ␥ 5 其 ⫽0. The propagator K N⫺1 is free of
O(a) corrections and, although otherwise not well behaved where
and so not useful in practice, it turns out to be very useful for
1
the construction of operators that are improved. In fact, the O ⌫⫾ ⫽ 关 O ⌫ A ⌫ B ⫾O ⌫F ⌫ 兴 , 共37兲
A⌫B 2 A B
four-fermion correlation function

冓 1
KN
⌫A
1 1
KN KN
⌫B 冔
1
KN and T̃⫽ ␴ ␮ ␯ ␥ 5 . Since the lattice preserves parity, the two

冓 冉 冊 冉 冊
sets of operators 共35兲 and 共36兲 renormalize independently.
1 1 1 1 1 Using continuous chiral transformations it is easy to show
⫽ 1⫺ aD N ⌫ A 1⫺ aD N
DN 2␳ 2␳ DN DN that, at variance with Wilson fermions, they renormalize

冉 冊 冉 冊 冔
1 1 1 plicitly verified this property at one loop in perturbation
⫻ 1⫺ aD N ⌫ B 1⫺ aD N 共34兲
2␳ 2␳ DN
is automatically improved. In the last line we have used Eq.
共33兲 to re-express the Green’s function in terms of the Neu-
berger propagator, and from this we can read off the im-
proved operator. Thus, the Ginsparg-Wilson relation highly
simplifies the improvement of the four-fermion operators at
all orders in perturbation theory. Moreover, the renormaliza-
tion factors for corresponding improved and unimproved op-
erators are the same. This happens because in 1-loop ampli-
tudes a D N factor can combine with a quark propagator but,
since it has an a in front and 共contrary to the Wilson case兲
there is no 1/a factor in the propagator 共as additive mass
renormalization is forbidden by chiral symmetry兲, and its
contribution to the renormalization factors is zero 关34兴.
The symmetries relevant for studying the mixing of the
operators that we consider are parity (P), the switch (S)
transformation ␺ 2 ↔ ␺ 4 and chiral symmetry in Eq. 共7兲. They
with the same renormalization matrices 关18兴. We have ex-
theory. In the following we will then consider only the
parity-conserving operators in Eqs. 共35兲. Among them, the
five Q ⫹i are left invariant under a switch transformation (S
⫽1), while the Q ⫺ i change sign (S⫽⫺1). Therefore the
two sets renormalize independently as
Q̂ ⫾ ⫾ ⫾
i 共 ␮ 兲 ⫽Ẑ i j 共 ␮ a 兲 Q j 共 a 兲 , 共38兲
where Ẑ ⫾ are the renormalization constant matrices.
The operators of the bases 共35兲 do not belong, in general,
to irreducible representations of the chiral group. Neverthe-
less, chiral symmetry imposes many constraints on the ma-
trix of the renormalization constants when a scheme which
preserves it together with the switch symmetry is chosen.
The most general mixing matrix under these constraints

冉 冊
reads
allow a splitting of the original basis into smaller indepen-
dent multiplets. For parity-conserving (P⫽1) operators it is ⫾
Z 11 0 0 0 0
useful to define the bases
0 Z 22 ⫾Z 23 0 0
Q⫾ ⫾ ⫾
1 ⫽O VV ⫹O AA 0 ⫾Z 32 Z 33 0 0
Ẑ ⫾ ⫽ . 共39兲
⫾ ⫾
Q⫾ ⫾ ⫾
2 ⫽O VV ⫺O AA
0 0 0 Z 44 Z 45
⫾ ⫾
0 0 0 Z 54 Z 55
Q⫾ ⫾ ⫾
3 ⫽O SS ⫺O P P

Q⫾ ⫾ ⫾
4 ⫽O SS ⫹O P P Note that for i, j⫽2,3, Z ⫹ ⫺
i j and Z i j are related. Therefore for
Neuberger fermions there are only 14 independent renormal-
Q⫾ ⫾
5 ⫽O TT 共35兲 ization constants, which would instead become 64 if the Wil-

114506-6
PERTURBATIVE RENORMALIZATION OF WEAK . . . PHYSICAL REVIEW D 62 114506

son action were used.5 The particular structure of the matrix

in Eq. 共39兲 also simplifies the implementation of non-
perturbative techniques 关33,39兴.
Analogously to the bilinears, denoting by x 1 ,x 3 and x 2 ,x 4
the coordinates of the outgoing and incoming quarks respec-
tively, the four-point Green’s functions are defined as

G⫾ ⫾
i 共 x 1 ,x 2 ,x 3 ,x 4 兲 ⫽ 具 ␺ 1 共 x 1 兲 ␺ 2 共 x 2 兲 Q i 共 0 兲 ␺ 3 共 x 3 兲 ␺ 4 共 x 4 兲 典 ,
¯ ¯
共40兲 FIG. 1. An example of Fierz and charge conjugation rearrange-
ments for Feynman graphs of four-fermion operators.
where 具 ••• 典 denotes the vacuum expectation value. Note
that G ⫾
i depends implicitly on the four color and Dirac indi- 23¯␥ (0)
S ⫹9 ␥ T
¯ (0) 41¯␥ (0)
S ⫺9 ␥ T
¯ (0)
⫹(0) ⫺(0)
ces carried by the external fermion fields. The Fourier trans- ␥ 44 ⫽ , ␥ 44 ⫽
12 12
form of the Green’s function 共40兲 at equal external momenta
p is defined as
⫹(0)
⫺ ¯␥ (0)
S ⫹␥T
¯ (0)
⫺(0)
5 共 ¯␥ (0)
S ⫺␥T 兲
¯ (0)

冕
␥ 45 ⫽ , ␥ 45 ⫽
12 12
G⫾
i 共 p 兲⫽ dx 1 dx 2 dx 3 dx 4 e ⫺ip(x 1 ⫹x 3 ⫺x 2 ⫺x 4 )
5 共 ¯␥ (0)
S ⫺␥T 兲
¯ (0) ⫺ ¯␥ (0)
S ⫹␥T
¯ (0)
⫻G ⫾
i 共 x 1 ,x 2 ,x 3 ,x 4 兲 . 共41兲 ⫹(0)
␥ 54 ⫽ , ⫺(0)
␥ 54 ⫽
4 4
The corresponding amputated correlation functions are de-
fined as
⫹(0)
⫺9 ¯␥ (0)
S ⫹41␥ T
¯ (0)
⫺(0)
9 ¯␥ (0)
S ⫹23␥ T
¯ (0)
␥ 55 ⫽ , ␥ 55 ⫽ ,
⌳⫾ ⫺1
共 p 兲 S ⫺1 共 p 兲 G ⫾ ⫺1
共 p 兲 S ⫺1 共 p 兲 . 12 12
i 共 p 兲 ⫽S i 共 p 兲S 共42兲
共45兲
From the above definitions we find, for the renormalized
which agree with 关37,38兴, where the anomalous dimension
Green’s functions,
matrices at two loops in the same bases can also be found.6
The renormalization group evolution of the four-fermion
Ĝ ⫾ 2 ⫾ ⫾
i 共 p 兲 ⫽Z ␺ Ẑ i j G j 共 p 兲 ,
operators, in a scheme which preserves chiral and the switch
ˆ⫾ ⫺2 ⫾ ⫾ symmetry, is
⌳ i 共 p 兲 ⫽Z ␺ Ẑ i j ⌳ j 共 p 兲 . 共43兲
Q̂ ⫾ 共 ␮ 兲 ⫽ŵ ⫾ 共 ␮ 兲关 ŵ ⫾ 共 ␮ 0 兲兴 ⫺1 Q̂ ⫾ 共 ␮ 0 兲 , 共46兲
The anomalous dimension matrices are defined as

冉 冊 where in this case ŵ ⫾ ( ␮ ) are matrices which depend only on

⬁
d ␣s i⫹1
ˆ⫾
␥ 共 ␣ s 兲 ⬅⫺2 共 Ẑ 兲 ⫾ ⫺1
␮ 2
d␮2
⫾
Ẑ ⫽ 兺␥ ˆ ⫾(i)
4␲
the anomalous dimensions and ␣ s . Their expression at NLO
i⫽0
in a generic scheme can be found in 关37,38,12兴. Analogously
共44兲 to the bilinears, the RGI operators can be defined as 关12兴
and, if the renormalization scheme preserves chiral symme-
try, they have the same structure as in Eq. 共39兲. At first order 关 Q̂ ⫾ 兴 RGI⫽ 关 ŵ ⫾ 共 ␮ 兲兴 ⫺1 Q̂ ⫾ 共 ␮ 兲 ⫽ 关 Ẑ ⫾ 共 a 兲兴 RGIQ ⫾ 共 a 兲 ,
in ␣ s we obtain 共47兲

2 4 where
⫹(0) ⫺(0)
␥ 11 ⫽⫺ ¯␥ (0) , ␥ 11 ⫽ ¯␥ (0)
3 S 3 S 关 Ẑ ⫾ 共 a 兲兴 RGI⫽ 关 ŵ ⫾ 共 ␮ 兲兴 ⫺1 Ẑ ⫾ 共 ␮ a 兲 . 共48兲
1 As for the scheme, scale and gauge dependences, these defi-
␥ 22
(0)
⫽⫺ ¯␥ (0) , ␥ 23
(0)
⫽⫺2 ¯␥ (0)
3 S S
nitions have the same properties as for bilinears.
We have computed the amputated Green’s functions of
8 the four-fermion operators at one loop in two different ways.
␥ 32
(0)
⫽0, ␥ 33
(0)
⫽ ¯␥ (0) In the first one we use the Fierz and charge conjugation
3 S
rearrangements 共see Fig. 1兲 to connect the proper vertices of
the four-fermion operators to the bilinear ones. In the second,
5
the calculation has been performed computing the four-quark
In the Wilson case the renormalization matrices of parity- diagrams directly without any rearrangement of quark legs.
conserving and parity-violating operators are different 关18兴. More- To carry out the analytic calculations we have used a set of
over, all the entries of the analogue of the matrices in Eq. 共39兲 are
independent and non-zero. This results in 25⫻2⫹14⫽64 indepen-
dent constants. It is interesting to note that using suitable Ward
identities, some parity-conserving matrix elements can be related to 6
The operators Q ⫾
5 used in this paper and defined in Eq. 共35兲
the parity-violating ones 关46兴. correspond to 1/4 of those defined in 关37,38兴.

114506-7
STEFANO CAPITANI AND LEONARDO GIUSTI PHYSICAL REVIEW D 62 114506

routines written in the symbolic manipulation language

FORM, which are a generalization to the overlap case of the
ones used in Refs. 关35兴. Many of the features of those calcu-
lations are also present here. We have numerically integrated
the analytic FORM outputs for some values of the ␳ parameter
and we have successfully checked the results obtained with
the first method. This also turns out to be a non-trivial check
of the numerical integrations relevant in the Neuberger regu-
larization. The full expressions for the one-loop amputated
Green’s functions that we have obtained in a generic covari-
ant gauge are reported in Appendix D, and they can be used
to impose any set of renormalization conditions.
Once the results have been rotated in the basis 共35兲, the RI
renormalization conditions are imposed by taking the trace
of the amputated Green’s functions in the Landau gauge
关18兴:
TrP⫾ ˆ⫾
i ⌳ k ⫽ ␦ ik 共 i,k⫽1, . . . ,5兲 ,
where the projectors P⫾ i are defined in Appendix C. Using
the wave function renormalization Z ␺RI defined in Eq. 共25兲
and the definition in 共43兲 we obtain
⫹
⫺9B S ⫹41B T ⫺
9B S ⫹23B T
B 55 ⫽ , B 55 ⫽ ,
12 12
共52兲
where the B ⌫ ’s are reported in Table I. The relations above
are valid in the Landau gauge for any action which preserves
chiral symmetry, where B A ⫽B V and B P ⫽B S .
The Wilson coefficients and the four-fermion operators of
the weak effective Hamiltonian are often defined in one of
the MS schemes. Unfortunately the definition of the MS
scheme for the four-fermion operators is not unique even if
we consider naive dimensional regularization only. This is a
source of confusion in the literature, where quite often one
encounters comparisons of matrix elements that have been
computed in different schemes. Incidentally, we note that the
MS-NDR scheme used in the lattice calculation of Ref. 关8兴
differs from the one used in 关37兴, which moreover is not the
共49兲 one adopted in 关38兴. In some cases, the differences between
various schemes may be numerically large. On the other
hand, the matrix elements of the four-fermion operators in a
given scheme are useful only if they are matched with the
Wilson coefficients computed in the same scheme. From the

冋 册
perturbative computation of the Wilson coefficients in a
g 20 ␥ˆ (0) given scheme ␹ , it is straightforward to extract the evolution
关 Ẑ ⫾ 兴 RI⫽1⫺ log共 ␮ a 兲 2 ⫹B̂ ⫾ ⫹2C F B ␺ ,
16␲ 2 2 operator 关 ŵ ⫾ ( ␮ ) 兴 ␹ at NLO. Then the corresponding renor-
共50兲 malization constants are given by

冉 冊
where 关 Ẑ ⫾ 共 ␮ a 兲兴 ␹ ⫽ 关 ŵ ⫾ 共 ␮ 兲兴 ␹ 关 Ẑ ⫾ 共 a 兲兴 RGI
⫾
B 11 0 0 0 0 ⫽ 关 ŵ ⫾ 共 ␮ 兲兴 ␹ 关 ŵ ⫾ 共 ␮ 兲兴 ⫺1 关 Ẑ ⫾ 共 ␮ a 兲兴 RI,
RI

0 B 22 ⫾B 23 0 0 共53兲
0 ⫾B 32 B 33 0 0
B̂ ⫾ ⫽ 共51兲 where 关 ŵ ⫾ ( ␮ ) 兴 RI can be found in 关37,12,38兴. This is equiva-
⫾ ⫾
0 0 0 B 44 B 45 lent to using only RGI Wilson coefficients and operators as
0 0 0 ⫾
B 54 ⫾
B 55 proposed in 关12兴.

V. CONCLUSIONS
and the B ⫾
i j are
The matrix elements of dimension-six four-fermion op-
⫹
⫺2 共 B S ⫺5B V 兲 ⫺
4 共 B S ⫹B V 兲 erators without power subtractions are the primary ingredi-
B 11 ⫽ , B 11 ⫽ ents for studying many interesting phenomenological prob-
3 3
lems in weak interactions, among which the most important
BS are the predictions of the K 0 -K̄ 0 and B 0 -B̄ 0 mixings in the
B 22⫽⫺ ⫹3B V , B 23⫽⫺2 共 B S ⫺B V 兲 standard model and beyond. The renormalization constants
3
of these operators are also relevant for studying the ⌬I
⫽1/2 rule and computing the C P-violation parameter ⑀ ⬘ / ⑀ .
8B S
B 32⫽0, B 33⫽ In this paper we have studied in detail the mixing of these
3 operators using the overlap lattice regularization proposed by
Neuberger and computed their renormalization factors at one
⫹
23B S ⫹9B T ⫺
41B S ⫺9B T loop in perturbation theory. The computations were done in
B 44 ⫽ , B 44 ⫽ two independent ways: in one case, using Fierz and charge
12 12
conjugation rearrangements, the four-fermion Green’s func-
⫺B S ⫹B T 5 共 B S ⫺B T 兲 tions are given in terms of the one loop results for the bilin-
⫹ ⫺
B 45 ⫽ , B 45 ⫽ ear operators; in the other case the Feynman diagrams are
12 12 computed directly using FORM codes.
We have shown that operators belonging to different chi-
⫹
5 共 B S ⫺B T 兲 ⫺
⫺B S ⫹B T ral representations do not mix among themselves. We have
B 54 ⫽ , B 54 ⫽
4 4 explicitly verified that, as expected, the mixing matrices for

114506-8
PERTURBATIVE RENORMALIZATION OF WEAK . . . PHYSICAL REVIEW D 62 114506

the parity-violating and parity-conserving operators are the ␳ ar

same. Furthermore, the improvement of the matrix elements X 0 共 p 兲 ⫽i ␥ ␮ p̄ ␮ ⫺ ⫹ p̂ 2
a 2
is highly simplified by the Ginsparg-Wilson relation and the
renormalization constants are the same as the unimproved
case. In the Wilson formulation, whether improved or not,
the construction of renormalized operators requires subtrac-
X 1 ␮ 共 p⫹ p ⬘ 兲 ⫽⫺ig 0 ␥ ␮ cos 冋 冉 冊 p ␮ ⫹ p ␮⬘
2
a

tions of operators with wrong chirality. These become very

severe in the case of power divergences, i.e. ⌬I⫽1/2, ⑀ ⬘ / ⑀ ,
etc. We believe that the overlap regularization represents a
⫺ir sin 冉 冊册 p ␮ ⫹ p ␮⬘
2
a

very promising approach for solving these long-standing

problems. X 2 ␮ ␯ 共 p⫹ p ⬘ 兲 ⫽⫺a
g 20
2
冋 冉 冊
␦ ␮ ␯ r cos
p ␮ ⫹ p ␮⬘
2
a

ACKNOWLEDGMENTS

We thank G. Martinelli for a very illuminating discussion

⫺i ␥ ␮ sin 冉 冊册 p ␮ ⫹ p ␮⬘
2
a . 共A3兲
on the chiral properties of the operators. L.G. thanks also C.
Hoelbling, V. Lubicz, C. Rebbi and M. Schwetz for stimu- The fermionic propagator of the overlap action is
lating discussions. S.C. has been supported in part by the
U.S. Department of Energy 共DOE兲 under cooperative re-
search agreement DE-FC02-94ER40818. L.G. has been sup-
S N(0) 共 p 兲 ⫽
a
冋 ⫺i ␥ ␳ p̄ ␳
2 ␳ ␻ 共 p 兲 ⫹b 共 p 兲
⫹1 ⫽ 册
a X †0 共 p 兲 ⫹ ␻ 共 p 兲
2 ␳ ␻ 共 p 兲 ⫹b 共 p 兲
,

ported in part under DOE grant DE-FG02-91ER40676. 共A4兲

with
APPENDIX A: FEYNMAN RULES
ar 2 ␳
In this appendix we report the Feynman rules used in the b共 p 兲⫽ p̂ ⫺ , 共A5兲
2 a
computations. Some abbreviations of functions occurring on
the lattice are useful:
冑 冉 ra 2 ␳
冊 2

冉 冊 ␻共 p 兲⫽ p̄ 2 ⫹ p̂ ⫺ 共A6兲
3 .
2 k ␮a
k̂ ␮ ⫽ sin
a 2
, k̂ 2 ⫽ 兺0 k̂ ␮2
2 a
For 0⬍ ␳ ⬍2r the propagator S N(0) (p) exhibits a massless
3 pole only when p ␮ ⫽0, and there are no doublers.
1
k̄ ␮ ⫽ sin共 k ␮ a 兲 , k̄ 2 ⫽ 兺0 k̄ ␮2 . 共A1兲 The vertices of the overlap Dirac operator are

再
a
␳ X 0共 p ⬘ 兲
The gluonic propagator in a generic covariant gauge is de- V ␮(1) 共 p,p ⬘ 兲 ⫽ X 1 ␮ 共 p⫹ p ⬘ 兲 ⫺
fined as ␻共 p⬘兲⫹␻共 p 兲 ␻共 p⬘兲
G ␮ ␯ 共 k 兲 ⫽ 关 ␦ ␮ ␯ ⫹ 共 ␣ ⫺1 兲 k̂ ␮ k̂ ␯ /k̂ 兴 /k̂ , 共A2兲
冎
2 2
X 0共 p 兲
⫻X 1† ␮ 共 p⫹ p ⬘ 兲 共A7兲
where ␣ is the gauge parameter. ␻共 p 兲
Let us define some useful matrices in Dirac space: and

V ␮(2)␯ 共 p, p ⬘ 兲 ⫽ 冋 ␳
␻共 p⬘兲⫹␻共 p 兲
再 X 2 ␮ ␯ 共 p⫹ p ⬘ 兲 ⫺
X 0共 p ⬘ 兲
␻共 p⬘兲
X 2† ␮ ␯ 共 p⫹ p ⬘ 兲
X 0共 p 兲
␻共 p 兲 冎
␳
⫹
2 兵 ␻ 共 p ⬘ 兲 ⫹ ␻ 共 p 兲 其兵 ␻ 共 p 兲 ⫹ ␻ 共 p⫹k 兲 其兵 ␻ 共 p⫹k 兲 ⫹ ␻ 共 p ⬘ 兲 其

再
⫻ X 0 共 p ⬘ 兲 X 1† ␮ 共 p⫹ p ⬘ ⫹k 兲 X 1 ␯ 共 2 p⫹k 兲 ⫹X 1 ␮ 共 p⫹ p ⬘ ⫹k 兲 X †0 共 p⫹k 兲 X 1 ␯ 共 2 p⫹k 兲

␻ 共 p ⬘ 兲 ⫹ ␻ 共 p 兲 ⫹ ␻ 共 p⫹k 兲
⫹X 1 ␮ 共 p⫹ p ⬘ ⫹k 兲 X †1 ␯ 共 2 p⫹k 兲 X 0 共 p 兲 ⫺
␻ 共 p ⬘ 兲 ␻ 共 p 兲 ␻ 共 p⫹k 兲

⫻X 0 共 p ⬘ 兲 X 1† ␮ 共 p⫹ p ⬘ ⫹k 兲 X 0 共 p⫹k 兲 X 1† ␯ 共 2 p⫹k 兲 X 0 共 p 兲 冎册 . 共A8兲

114506-9
STEFANO CAPITANI AND LEONARDO GIUSTI PHYSICAL REVIEW D 62 114506

APPENDIX B: NOTATIONS AND FIERZ
TRANSFORMATIONS
In this appendix we fix our notation for the color and spin
matrices. We also report the formulas used for Fierz trans-
formations in color and Dirac space.
The Gell-Mann group generators of the SU(N c ) Lie alge-
The Fierz transformation of the Dirac indices of a four-
fermion operator is defined as
⌫ 丢 ⌫⬅⌫ ␣␤ 丢 ⌫ ␥ ␦ → 关 ⌫ 丢 ⌫ 兴 F D ⬅⌫ ␣ ␦ 丢 ⌫ ␥␤ . 共B10兲
The Euclidean Fierz-transformed Dirac tensor products 关 ⌫
丢 ⌫ 兴 F D can be re-expressed as a linear combination of the

冉 冊冉
bra are denoted by t A , A⫽1, . . . ,N 2c ⫺1. They are Hermitian, complete set of the original tensor products ⌫ 丢 ⌫ as follows:

冊
traceless N c ⫻N c matrices and are normalized according to
关 S 丢 S 兴 FD 1 1 ⫺1 ⫺1 1
1
Tr共 t A t B 兲 ⫽ ␦ AB . 共B1兲 关 V 丢 V 兴 FD
2 4 ⫺2 0 ⫺2 ⫺4
关 T 丢 T 兴 FD 1
⫽⫺ ⫺6 0 ⫺2 0 ⫺6
They satisfy the commutation relations 4
关 A 丢 A 兴 FD ⫺4 ⫺2 0 ⫺2 4
关 t ,t 兴 ⫽i f
A B ABC C
t , 共B2兲

冉冊
关P 丢 P兴 FD
1 ⫺1 ⫺1 1 1
where summation over repeated indices is implied. The
structure constants f ABC are completely antisymmetric and S丢S
real. With these conventions the completeness relation for V丢V
the t generators reads
⫻ T丢T . 共B11兲

冉 冊
2
N c ⫺1 A丢A
1 1
兺
A⫽1
共 t A 兲 ab 共 t A 兲 cd ⫽ ␦ ␦ ⫺ ␦ ␦ ,
2 ad bc N c ab cd
共B3兲 P丢P

and using the above formulas we get

The overall minus sign is due to the anti-commutativity of
2
N c ⫺1 the Fermi fields.
兺
A⫽1
共 t A t A 兲 ab ⫽C F ␦ ab , 共B4兲
APPENDIX C: PROJECTORS FOR BILINEARS AND
2 FOUR-FERMION OPERATORS
N c ⫺1

兺
C,D⫽1
f ACD f BCD ⫽C A ␦ AB , 共B5兲 In this appendix we define the projectors used to impose
the RI renormalization conditions for the bilinear and four-
fermion operators.
with C F ⫽(N 2c ⫺1)/2N c and C A ⫽N c . For bilinears the projectors are defined as
The complete basis of 16 Euclidean Dirac 4⫻4 matrices
is denoted by 1 1
PS ⫽ I, PP ⫽ ␥
⌫⫽ 兵 1, ␥ ␮ , ␴ ␮ ␯ , ␥ ␮ ␥ 5 , ␥ 5 其 ⬅ 兵 S,V,T,A, P 其 , 共B6兲 12 12 5

where ␥ i are the usual Euclidean Dirac matrices in four di- 1 1

PV ⫽ ␥ , PA ⫽ ␥ ␥
mensions and 48 ␮ 48 5 ␮
1
␴ ␮␯⫽ 关 ␥ ␮ , ␥ ␯兴 . 共B7兲 1
2 PT ⫽⫺ ␴ , 共C1兲
72 ␮ ␯
The charge conjugation matrix satisfies
where the identity color matrix is not shown.
C ␥ ␮T C ⫺1 ⫽⫺ ␥ ␮ , 共B8兲 If we define

and in our basis C⫽ ␥ 0 ␥ 2 . P⌫⌫ ⬅⌫ 丢 ⌫, 共C2兲

Repeated ⌫ matrices imply summation of their Lorenz
indices 共if any兲; for example VV⫽ 兺 ␮ ␥ ␮ 丢 ␥ ␮ , TT its trace on a generic four-fermion amputated Green’s func-
⫽ 兺 ␮ ⬍ ␯ ␴ ␮ ␯ 丢 ␴ ␮ ␯ , where the sum is over the 6 independent tion is defined as 关18兴
␴ ␮ ␯ matrices only. With these conventions the ⌫ matrices
TrP⌫⌫ ⌳ ⌫⌫ 共 p 兲 ⫽⌫ ␴␳ 丢 ⌫ ␴ ⬘ ␳ ⬘ ⌳ ⌫⌫ 共 p 兲 ␳␴␳ ⬘⬘␴ ⬘⬘ ,
are normalized as RRR R
共C3兲
SS⫽1, VV⫽4, TT⫽⫺6, AA⫽⫺4, P P⫽1,
where the color and spinor indices are explicitly reported and
共B9兲
the trace is taken over spin and color. The projectors P⫾k for
where summation over Dirac indices is understood. the parity conserving operators Q ⫾
k are 关18兴

114506-10
PERTURBATIVE RENORMALIZATION OF WEAK . . . PHYSICAL REVIEW D 62 114506

TABLE II. Values of the finite constants in Landau gauge.

␳ ⫽0.4 ␳ ⫽1.0 ␳ ⫽1.7

S ⫺1 ⫽ip” 1⫹ 再 g 20
16␲ 2
C F 关 ␣ L⫹B ␺ ⫹ ␣ d ␺ 兴 , 冎 共D1兲

c1 4.761193 8.939783 12.630887

c2 ⫺0.142685 ⫺0.488442 ⫺0.790006 and the amputated Green’s functions of the bilinear operators
c3 ⫺1.524484 ⫺1.539241 ⫺1.575409 in a generic covariant gauge are
c4 3.048969 3.078482 3.150818

再 冎
c5 ⫺0.428056 ⫺1.465325 ⫺2.370017
c6 ⫺1.952540 ⫺3.004566 ⫺3.945426 g 20
⌳ S ⫽S 1⫹ C F 关 ⫺ 共 3⫹ ␣ 兲 L⫹B S ⫺ ␣ 共 d ␺ ⫺1 兲兴
c7 2.478227 1.124716 ⫺0.009205 16␲ 2
c8 ⫺0.856112 ⫺2.930650 ⫺4.740034

再 冎
c9 ⫺2.095225 ⫺3.493007 ⫺4.735432
g 20
⌳ V ⫽V 1⫹ C F 关 ⫺ ␣ L⫹B V ⫺ ␣ d ␺ 兴
1 16␲ 2
P⫾
1 ⬅⫹ 共 P ⫹PAA 兲
64N c 共 N c ⫾1 兲 VV g 20 p ␮ p”
⫺2 C F␣
16␲ 2
p2
1
P⫾
2 ⬅⫹ 共 PVV ⫺PAA 兲

再 冎
64共 N 2c ⫺1 兲
g 20
1 ⌳ T ⫽T 1⫹ C F 关共 1⫺ ␣ 兲 L⫹B T ⫺ ␣ 共 d ␺ ⫹1 兲兴
⫾ 共 PSS ⫺PP P 兲 16␲ 2
32N c 共 N 2c ⫺1 兲

P⫾
3 ⬅⫾
1
32N c 共 N 2c ⫺1 兲
共 PVV ⫺PAA 兲 再
⌳ A ⫽A 1⫹
g 20
16␲ 2
C F 关 ⫺ ␣ L⫹B A ⫺ ␣ d ␺ 兴 冎
1 g 20 p ␮ p” ␥ 5
⫹ 共 PSS ⫺PP P 兲 ⫺2 C F␣
16共 N 2c ⫺1 兲 16␲ 2
p2

再 冎
共 2N c ⫾1 兲
P⫾
4 ⬅⫹ 共 PSS ⫹PP P 兲 g 20
32N c 共 N 2c ⫺1 兲 ⌳ P ⫽ P 1⫹ C F 关 ⫺ 共 3⫹ ␣ 兲 L⫹B P ⫺ ␣ 共 d ␺ ⫺1 兲兴 ,
16␲ 2
1 共D2兲
⫿ PTT
32N c 共 N 2c ⫺1 兲
with L⫽log(pa)2, and the B ⌫ s are reported in Table I. The
1 terms proportional to the gauge parameter ␣ can be given in
P⫾
5 ⬅⫿ 共 PSS ⫹PP P 兲 simple form as functions of d ␺ ⫽ ␥ E ⫺F 0 ⫺1⫽⫺4.792010
32N c 共 N 2c ⫺1 兲 共where ␥ E is the Euler constant and F 0 ⫽4.3692252 . . . ).
They depend on the gluonic action, and they are independent
共 2N c ⫿1 兲
⫹ PTT . 共C4兲 of the ␳ parameter 关42兴, as can be seen using gauge Ward
96N c 共 N 2c ⫺1 兲 Identities.
The four-fermion operators that we have considered in the
They are obtained from the tree-level amputated Green’s ‘‘direct’’ calculation have the form
functions ⌳ ⫾(0)
k imposing the following orthogonality condi-
tions:
O ⌫ A ⌫ B ⫽ ¯␺ 1 ⌫ A ␺ 2 • ¯␺ 3 ⌫ B ␺ 4
TrP⫾ ⫾(0)
i ⌳k ⫽ ␦ ik 共 i,k⫽1, . . . ,5兲 . 共C5兲

The analogous projectors for the parity-violating operators O ⌫ c ⌫ c ⫽ ¯␺ 1 ⌫ A t C ␺ 2 • ¯␺ 3 ⌫ B t C ␺ 4 .

A B
can be found in 关18兴. 共D3兲

APPENDIX D: ONE-LOOP RESULTS

The one-loop expressions for the amputated Green’s func-
The one-loop expression for the quark propagator in a tions of the parity-conserving operators in a general covari-
generic covariant gauge is ant gauge are

114506-11
STEFANO CAPITANI AND LEONARDO GIUSTI PHYSICAL REVIEW D 62 114506

⌳ S c S c ⫽O S c S c ⫹
g 20
16␲ 2
冠再 ⫺
1
2N c 冋
关 ⫺2 共 3⫹ ␣ 兲 L⫹c 1 ⫺2 ␣ 共 d ␺ ⫺1 兲兴 ⫺N c ␣ L⫹c 3 ⫹ ␣ d ␺ ⫹
1
2 冉 冊 册冎 O ScSc

⫹
N 2c ⫺1
N 2c
冋冉 1
2
1
冊 p␣p␤
L⫹c 2 O TT ⫺ ␣ ␴ ␳ ␣ 丢 ␴ ␳ ␤ 2 ⫹
2 p
N 2c ⫺4
Nc 册 冋冉 1
2 冊 1
2
p␣p␤
L⫹c 2 O T c T c ⫺ ␣ ␴ ␳ ␣ t A 丢 ␴ ␳ ␤ t A 2
p
册冡
⌳ SS ⫽O SS ⫹
16␲
g 20
2
冠 N 2c ⫺1
2N c
关 ⫺2 共 3⫹ ␣ 兲 L⫹c 1 ⫺2 ␣ 共 d ␺ ⫺1 兲兴 O SS ⫹4 冋冉 1
2 冊 1 p␣p␤
L⫹c 2 O T c T c ⫺ ␣ ␴ ␳ ␣ t A 丢 ␴ ␳ ␤ t A 2
2 p
册冡
⌳ P c P c ⫽O P c P c ⫹
g 20
16␲ 2
冠再 ⫺
1
2N c 冋
关 ⫺2 共 3⫹ ␣ 兲 L⫹c 1 ⫺2 ␣ 共 d ␺ ⫺1 兲兴 ⫺N c ␣ L⫹c 3 ⫹ ␣ d ␺ ⫹
1
2 冉 冊 册冎 O PcPc

⫹
N 2c ⫺1
N 2c
冋冉 1
2
1 1
冊 p␣p␤
L⫹c 2 ⫺ ␣ O TT ⫹ ␣ ␴ ␳ ␣ 丢 ␴ ␳ ␤ 2 ⫹
2 2 p
N 2c ⫺4
Nc 册 冋冉 1
2
1
L⫹c 2 ⫺ ␣ O T c T c
2 冊
1 p␣p␤
⫹ ␣ ␴ ␳␣t A 丢 ␴ ␳␤t A 2
2 p
册冡
⌳ P P ⫽O P P ⫹
16␲
g 20
2 再 N 2c ⫺1
2N c
关 ⫺2 共 3⫹ ␣ 兲 L⫹c 1 ⫺2 ␣ 共 d ␺ ⫺1 兲兴 O P P ⫹4 冋冉 1
2
1
L⫹c 2 ⫺ ␣ O T c T c
2 冊
1 p␣p␤
⫹ ␣ ␴ ␳␣t A 丢 ␴ ␳␤t A 2
2 p
册冎
⌳ V c V c ⫽O V c V c ⫹
g 20
16␲ 2
冠再 ⫺
1
2N c
共 ⫺2 ␣ L⫹c 4 ⫺2 ␣ d ␺ 兲 ⫺N c 冋冉 冊 3
2
⫹ ␣ L⫹c 6 ⫹ ␣ d ␺ 冊 册冎 O VcVc⫹ 冉 2
Nc
⫹N c ␣ 冊
p” t A 丢 p” t A
p2

⫹
N 2c ⫺1
N 2c
冋冉 3
2
L⫹c 5 ⫺ ␣ O AA ⫹ ␣ 冊
p” ␥ 5 丢 p” ␥ 5
p2
⫹
N 2c ⫺4
Nc 册 冋冉 3
2
L⫹c 5 ⫺ ␣ O A c A c ⫹ ␣冊p” ␥ 5 t A 丢 p” t A ␥ 5
p2
册
⌳ VV ⫽O VV ⫹
g 20
16␲ 2
再 冉 N 2c ⫺1
2N c
共 ⫺2 ␣ L⫹c 4 ⫺2 ␣ d ␺ 兲 O VV ⫺4 ␣
p” 丢 p”
p2
冊 冋冉⫹4
3
2 冊
L⫹c 5 ⫺ ␣ O A c A c ⫹ ␣
p” ␥ 5 t A 丢 p” ␥ 5 t A
p2
册冎
⌳ A c A c ⫽O A c A c ⫹
16␲
g 20
2
冠再 ⫺
1
2N c
共 ⫺2 ␣ L⫹c 4 ⫺2 ␣ d ␺ 兲 ⫺N c 冋冉 冊
3
2
⫹ ␣ L⫹c 6 ⫹ ␣ d ␺ 册冎 O AcAc

⫹ 冉 2
Nc
⫹N c ␣ 冊
p” ␥ 5 t A 丢 p” ␥ 5 t A N 2c ⫺1
p2
⫹
N 2c
冋冉 3
2 冊 p” 丢 p”
L⫹c 5 ⫺ ␣ O VV ⫹ ␣ 2 ⫹
p
N 2c ⫺4
Nc 册 冋冉 3
2 冊
L⫹c 5 ⫺ ␣ O V c V c

⫹␣
p” t A 丢 p” t A
p2
册冡
⌳ AA ⫽O AA ⫹
g 20
16␲ 2
再 冉 N 2c ⫺1
2N c
共 ⫺2 ␣ L⫹c 4 ⫺2 ␣ d ␺ 兲 O AA ⫺4 ␣
p” ␥ 5 丢 p” ␥ 5
p2
冊 冋冉
⫹4
3
2 冊
L⫹c 5 ⫺ ␣ O V c V c ⫹ ␣
p” t A 丢 p” t A
p2
册冎
⌳ T c T c ⫽O T c T c ⫹
16␲
g 20
2
冠再 ⫺
1
2N c 冋
关 2 共 1⫺ ␣ 兲 L⫹c 7 ⫺2 ␣ 共 d ␺ ⫹1 兲兴 ⫺N c 共 2⫹ ␣ 兲 L⫹c 9 ⫹ ␣ d ␺ ⫺
1
2 冉 冊 册冎 O TcTc

⫹
N 2c ⫺1
N 2c
冉 3
3L⫹c 8 ⫺ ␣ 共 O SS ⫹O P P 兲 ⫹
2 冊
N 2c ⫺4
Nc
3
冉
3L⫹c 8 ⫺ ␣ 共 O S c S c ⫹O P c P c 兲
2 冊 冡
114506-12
PERTURBATIVE RENORMALIZATION OF WEAK . . . PHYSICAL REVIEW D 62 114506

⌳ TT ⫽O TT ⫹
g 20
16␲ 2 再 N 2c ⫺1
2N c 冉 3
冊
关 2 共 1⫺ ␣ 兲 L⫹c 7 ⫺2 ␣ 共 d ␺ ⫹1 兲兴 O TT ⫹4 3L⫹c 8 ⫺ ␣ 共 O S c S c ⫹O P c P c 兲 .
2 冎 共D4兲

The c i constants, which take the same values also for the mixings of the parity-violating operators 共as we have explicitly
verified兲, are reported in Table II. They correspond to the values of the finite parts in the Landau gauge and, in terms of the
finite parts of the bilinears, are:

B T ⫺B S B S ⫹3B T ⫹4B V
c 1 ⫽2B S , c 2⫽ , c 3 ⫽⫺
8 8

B V ⫺B S B S ⫹B V
c 4 ⫽2B V , c 5⫽ , c 6 ⫽⫺
2 2

3 1
c 7 ⫽2B T , c 8 ⫽⫺ 共 B S ⫺B T 兲 , c 9 ⫽⫺ 共 3B S ⫹B T 兲 . 共D5兲
4 4

The coefficients of the logarithms can also be re-expressed in terms of of the anomalous dimensions of the bilinears.

关1兴 N. Cabibbo, G. Martinelli, and R. Petronzio, Nucl. Phys. B244,
381 共1984兲; R. C. Brower, M. B. Gavela, R. Gupta, and G.
Maturana, Phys. Rev. Lett. 53, 1318 共1984兲; C. Bernard in
Proceedings of Gauge Theory on a Lattice, Argonne, 1984, p.
85.
关2兴 M. Bochicchio et al., Nucl. Phys. B262, 331 共1985兲.
关3兴 S. Sharpe et al., Nucl. Phys. B286, 253 共1987兲.
关4兴 G. Kilcup, S. Sharpe, R. Gupta, and A. Patel, Phys. Rev. Lett.
64, 25 共1990兲.
关5兴 S. R. Sharpe, Nucl. Phys. B 共Proc. Suppl.兲 53, 181 共1997兲; L.
Lellouch, hep-ph/9906497.
关6兴 C. Bernard et al., Nucl. Phys. B 共Proc. Suppl.兲 4, 483 共1988兲.
关7兴 E. Franco et al., Nucl. Phys. B317, 63 共1989兲; M. B. Gavela
et al., ibid. B306, 677 共1988兲.
关8兴 S. R. Sharpe and A. Patel, Nucl. Phys. B417, 307 共1994兲; R.
Gupta, T. Bhattacharya, and S. R. Sharpe, Phys. Rev. D 55,
4036 共1997兲.
关9兴 M. Crisafulli et al., Phys. Lett. B 369, 325 共1996兲; L. Conti
et al., ibid. 421, 273 共1998兲.
关10兴 JLQCD Collaboration, S. Aoki et al., Phys. Rev. D 60, 034511
共1999兲.
关11兴 C. R. Allton et al., Phys. Lett. B 453, 30 共1999兲.
关12兴 A. Donini, V. Giménez, L. Giusti, and G. Martinelli, Phys.
Lett. B 470, 233 共1999兲.
关13兴 M. Ciuchini et al., J. High Energy Phys. 10, 008 共1998兲.
关14兴 F. Parodi, P. Roudeau, and A. Stocchi, Nuovo Cimento A 112,
833 共1999兲; F. Caravaglios, F. Parodi, P. Roudeau, and A.
Stocchi, hep-ph/0002171.
关15兴 M. Ciuchini, E. Franco, L. Giusti, V. Lubicz, and G. Marti-
nelli, Nucl. Phys. B573, 201 共2000兲.
关16兴 L. Giusti, A. Romanino, and A. Strumia, Nucl. Phys. B550, 3
共1999兲; R. Barbieri, L. J. Hall, and A. Romanino, ibid. B551,
93 共1999兲.
关17兴 L. Maiani, G. Martinelli, G. C. Rossi, and M. Testa, Nucl.
Phys. B289, 505 共1987兲.
关18兴 A. Donini, V. Giménez, G. Martinelli, M. Talevi, and A.
Vladikas, Eur. Phys. J. C 10, 121 共1999兲.
关19兴 P. Hasenfratz, Nucl. Phys. B 共Proc. Suppl.兲 63A-C, 53 共1998兲.
关20兴 P. Hasenfratz, V. Laliena, and F. Niedermayer, Phys. Lett. B
427, 125 共1998兲.
关21兴 H. Neuberger, Phys. Lett. B 417, 141 共1998兲; 427, 353 共1998兲.
关22兴 P. Hasenfratz, Nucl. Phys. B525, 401 共1998兲.
关23兴 M. Lüscher, Phys. Lett. B 428, 342 共1998兲.
关24兴 P. H. Ginsparg and K. G. Wilson, Phys. Rev. D 25, 2649
共1982兲.
关25兴 T. Chiu, C. Wang, and S. V. Zenkin, Phys. Lett. B 438, 321
共1998兲.
关26兴 D. Kaplan, Phys. Lett. B 288, 342 共1992兲; Y. Shamir, Nucl.
Phys. B406, 90 共1993兲; Phys. Rev. D 62, 054513 共2000兲.
关27兴 P. Hernández, K. Jansen, and M. Lüscher, Nucl. Phys. B552,
363 共1999兲.
关28兴 S. Capitani, M. Göckeler, R. Horsley, P. E. L. Rakow, and G.
Schierholz, Phys. Lett. B 468, 150 共1999兲; S. Capitani, M.
Göckeler, R. Horsley, P. E. L. Rakow, and G. Schierholz,
Nucl. Phys. B 共Proc. Suppl.兲 83-84, 893 共2000兲; S. Capitani
et al., hep-lat/0007004.
关29兴 H. Neuberger, Phys. Rev. Lett. 81, 4060 共1998兲; R. G. Ed-
wards, U. M. Heller, and R. Narayanan, Nucl. Phys. B540, 457
共1999兲; A. Bode, U. M. Heller, R. G. Edwards, and R. Naray-
anan, hep-lat/9912043; P. Hernández, K. Jansen, and L. Lel-
louch, hep-lat/0001008; R. Narayanan and H. Neuberger,
Phys. Rev. D 62, 074504 共2000兲; S. J. Dong, F. X. Lee, K. F.
Liu, and J. B. Zhang, hep-lat/0006004; L. Giusti, C. Hoelbling,
and C. Rebbi, Nucl. Phys. B 共Proc. Suppl.兲 83-84, 896 共2000兲;
and 共in preparation兲.
关30兴 M. Beneke, G. Buchalla, and I. Dunietz, Phys. Rev. D 54,
4419 共1996兲 and references therein.
关31兴 C. Dawson et al., Nucl. Phys. B514, 313 共1998兲; M. Testa, J.
High Energy Phys. 04, 002 共1998兲.
关32兴 M. Ciuchini, E. Franco, G. Martinelli, L. Reina, and L. Silves-
trini, Z. Phys. C 68, 239 共1995兲.
关33兴 G. Martinelli, C. Pittori, C. T. Sachrajda, M. Testa, and A.
Vladikas, Nucl. Phys. B445, 81 共1995兲.

114506-13
STEFANO CAPITANI AND LEONARDO GIUSTI
关34兴 C. Alexandrou, E. Follana, H. Panagopoulos, and E. Vicari,
Nucl. Phys. B580, 394 共2000兲.
关35兴 S. Capitani et al., Nucl. Phys. B570, 393 共2000兲; Nucl. Phys. B
共Proc. Suppl.兲 83, 232 共2000兲.
关36兴 G. Martinelli, Phys. Lett. 141B, 395 共1984兲.
关37兴 M. Ciuchini et al., Nucl. Phys. B523, 501 共1998兲.
关38兴 A. J. Buras, M. Misiak, and J. Urban, hep-ph/0005183.
关39兴 M. Lüscher, S. Sint, R. Sommer, and H. Wittig, Nucl. Phys.
B491, 344 共1997兲.
关40兴 F. Niedermayer, Nucl. Phys. B 共Proc. Suppl.兲 73, 105 共1999兲.
114506-14
PHYSICAL REVIEW D 62 114506
关41兴 M. Ishibashi, Y. Kikukawa, T. Noguchi, and A. Yamada, Nucl.
Phys. B576, 501 共2000兲.
关42兴 S. Capitani, hep-lat/0005008.
关43兴 V. Gimenez, L. Giusti, F. Rapuano, and M. Talevi, Nucl. Phys.
B540, 472 共1999兲.
关44兴 A. González-Arroyo, F. J. Yndurain, and G. Martinelli, Phys.
Lett. 117B, 437 共1982兲; 122B, 486E 共1983兲.
关45兴 S. Capitani, M. Lüscher, R. Sommer, and H. Wittig, Nucl.
Phys. B544, 669 共1999兲.
关46兴 D. Becirevic, et al., Phys. Lett. B 487, 74 共2000兲.